gallegon / capstone-tsfapc

CS461 OSU Capstone Tree Segmentation from Aerial Point Clouds

Capstone-TSFAPC

CS461 OSU Capstone Tree Segmentation from Aerial Point Clouds

GitHub repo: https://github.com/gallegon/Capstone-TSFAPC

Created by Mark G, Nicholai G, and Samuel F.

Installation and Usage

Basic requirements: Python, numpy, scipy, gdal, pdal.

To see full documentation of installation and usage please visit our GitHub pages site at: https://gallegon.github.io/Capstone-TSFAPC/

To run the algorithm use the provided ts_cli.py script:

python ts_cli.py context_file

Where context_file is the path to the input context JSON file. There are example JSON config files in the tests/ directory. These files allow the user to specify parameters for the tree segmentation algorithm.

Alternatively, the treesegmentation.treeseg_lib module can be imported and used in a custom Python script.

List of unrealized features

There are several features that we were unable to achieve during in our Capstone timeline:

• Integrating statistics processing into our script.
• Full integration with PDAL pipeline.
• Full integration as a QGIS plugin.
• Optimizations for hierarchy building, weighted graph, and partitioning steps.

Potential for future projects

A main goal of this project was to provide an extensible base for this tree segmentation algorithm.

• Further optimization of the hierachy building, weighted graph, and partitioning steps.
• Further extending the QGIS integration.
• Building the algorithm as a filter and/or writer for PDAL (in C++), reducing the need for Python to be used for processing. Using python as a front end to an interface (command line or GUI in QGIS) and leaving the heavy lifting of the algorithm to compiled code.
• More packaging for different platforms (Mac/Linux).